Methods and systems may provide for identifying, and distinguishing between electrical loads using time and frequency domain analysis of at least one property of an alternating current during a transient event. In one example, time and frequency domain features may be computed from the voltage signatures of an ON event. A support vector machine classifier may then be trained using the feature vectors (including the time and frequency domain features) for known devices. The trained support vector machine may classify or identify an unknown electrical device using a feature vector as input.
Legal claims defining the scope of protection. Each claim is shown in both the original legal language and a plain English translation.
1. A system, comprising: a sensor coupled to a leg in an electrical system, the sensor to generate samples corresponding to at least one property of an alternating current; a sensing module, in communication with the sensor, to capture samples of the at least one property of the alternating current during an ON event; and a computation module, in communication with the sensing module, to identify an electrical device using time domain and frequency analysis of the at least one property, wherein the computation module includes: a training module having a training database module and a support vector machine classifier, and a classification module having a trained support vector machine classifier and a classification database.
The system identifies electrical devices by analyzing the voltage and current characteristics when they turn on. A sensor monitors the alternating current in the electrical system. A sensing module captures data from the sensor during the "ON event" of a device. A computation module analyzes this data in both time and frequency domains to identify the device. This module includes a training section with a database and a support vector machine (SVM) classifier, used to learn patterns from known devices. The trained SVM then resides in the classification module with a classification database to identify unknown devices based on their "ON event" signatures.
2. The system of claim 1 , wherein the sensing module includes: a data acquisition module, a peak detect filtering module, and an event detection module.
The electrical device identification system as described previously is improved by specifying the components of the sensing module. The sensing module, which captures data from the sensor during the "ON event" of a device, contains a data acquisition module, a peak detect filtering module, and an event detection module. The data acquisition module initially obtains the raw signal. The peak detect filtering module then processes this signal. Finally, the event detection module pinpoints the exact moment the device turns on to trigger the data capture.
3. The system of claim 2 , wherein the data acquisition module is to capture an alternating current line voltage signal and perform an analog-to-digital conversion of the signal.
In the electrical device identification system previously described, the data acquisition module, part of the sensing module, specifically captures the alternating current line voltage signal and converts it from an analog signal to a digital signal using an analog-to-digital converter. This digital representation of the voltage signal is then used for further analysis by other modules within the system, such as the peak detect filtering module and the event detection module.
4. The system of claim 2 , wherein the peak detect filtering module is to compute a peak-to-peak voltage of the alternating current.
In the electrical device identification system, the peak detect filtering module, residing within the sensing module, calculates the peak-to-peak voltage of the alternating current. This value, representing the difference between the maximum and minimum voltage levels, is a characteristic feature of the electrical signal that can be used in identifying the specific electrical device turning on. This value contributes to the overall signature used by the classification module.
5. The system of claim 2 , wherein the event detection module is to detect the ON event of the electrical device.
The electrical device identification system utilizes an event detection module, part of the sensing module, to specifically detect the "ON event" of an electrical device. This module determines when a device is switched on, signaling the system to begin capturing and analyzing the electrical signature for device identification. Accurate and timely detection of the "ON event" is crucial for capturing relevant data for classification.
6. The system of claim 1 , wherein the computation module further includes a feature extraction module.
In the electrical device identification system, the computation module, responsible for identifying devices, also incorporates a feature extraction module. This module prepares the data for the machine learning component by computing relevant characteristics, or features, from the captured voltage and current data, making it easier for the SVM classifier to learn and accurately identify devices.
7. The system of claim 6 , wherein the feature extraction module is to compute a feature vector including frequency domain features and time domain features.
In the electrical device identification system featuring a feature extraction module, the feature extraction module computes a feature vector that contains both frequency domain features and time domain features. This combines information about how the signal changes over time with its frequency components. These extracted features are then used as input to the support vector machine classifier for training and device identification.
8. The system of claim 1 , wherein the training module is to train the support vector machine classifier.
The electrical device identification system includes a training module, as part of its computation module, specifically designed to train the support vector machine (SVM) classifier. This training process allows the SVM to learn the characteristic electrical signatures of different devices, enabling it to accurately identify unknown devices later during the classification phase.
9. The system of claim 8 , wherein the training module uses a database of feature vectors for known devices to perform the training.
Building upon the electrical device identification system, the training module uses a database of feature vectors for known devices to train the support vector machine (SVM) classifier. This database provides the SVM with examples of the electrical signatures of various devices, allowing it to learn the patterns associated with each device and improve its identification accuracy.
10. The system of claim 1 , wherein the classification module maintains a database to receive and store feature vectors for unknown devices.
Within the electrical device identification system's classification module, a database is maintained to receive and store feature vectors for unknown devices. This allows for logging and later analysis of unidentified devices or for improving the system's training data with new device signatures. This database acts as a repository for data collected during the classification process.
11. The system of claim 10 , wherein the classification module uses a trained support vector machine classifier to perform the identifying of unknown devices using a feature vector as input.
Continuing with the electrical device identification system, the classification module uses a trained support vector machine (SVM) classifier to identify unknown devices by using a feature vector as input. This means that after the SVM has been trained on known device signatures, it can then analyze the electrical signature of an unknown device and, based on its training, classify the device.
12. A method, comprising: capturing, via a sensing module, samples of at least one property of an alternating current during a transient event; identifying, via a computation module in communication with the sensing module, an electrical device using time and frequency domain analysis of the at least one property; training, via a training module having a training database module and a support vector machine classifier, a support vector machine classifier; and identifying, via a classification module having a trained support vector machine classifier and a classification database, an unknown electrical device.
A method is used to identify electrical devices based on their power-on characteristics. A sensing module captures samples of voltage and current during a device's "ON event." A computation module, linked to the sensing module, analyzes these samples in the time and frequency domains to pinpoint the device. A training module, with a training database and support vector machine (SVM) classifier, trains the SVM. The trained SVM then lives in a classification module, alongside a classification database, to classify unknown devices.
13. The method of claim 12 , further comprising: capturing, via a data acquisition module, a line voltage signal of the alternating current and performing an analog-to-digital conversion of the signal; computing, via a peak detect filtering module, a peak-to-peak voltage of the alternating current; and detecting, via an event detection module, the transient event of the electrical device.
The method of identifying electrical devices as described previously includes specific steps within the sensing module: A data acquisition module captures the alternating current line voltage signal and performs an analog-to-digital conversion. A peak detect filtering module computes the peak-to-peak voltage of the alternating current. An event detection module detects the transient event (ON event) of the electrical device, triggering the data capture and analysis process.
14. The method of claim 12 , further comprising: computing, via a feature extraction module, a feature vector including frequency domain features and time domain features.
The method of identifying electrical devices further includes the step of computing a feature vector using a feature extraction module. This feature vector contains both frequency domain features and time domain features extracted from the captured voltage and current data. This feature vector is subsequently used as input to the support vector machine classifier for device identification.
15. The method of claim 12 , wherein the training module uses a database of feature vectors for known devices to perform the training.
Within the described method for identifying electrical devices, the training of the support vector machine (SVM) classifier relies on a database of feature vectors from known devices. The training module uses this database to enable the SVM to recognize and classify different electrical devices based on their unique electrical signatures during the "ON event."
16. The method of claim 12 , wherein the classification module uses the trained support vector machine classifier to perform the identifying of unknown devices using a feature vector as input.
In the method for identifying electrical devices, the classification module uses the trained support vector machine (SVM) classifier to identify unknown devices using a feature vector as input. The trained SVM analyzes the extracted features and assigns the unknown device to the most likely category based on the patterns learned during the training phase.
17. At least one non-transitory computer readable storage medium comprising a set of instructions which, if executed by a computing system, cause the computing system to: capture samples of at least one property of an alternating current during an ON event, identify an electrical device using time domain and frequency domain analysis of the at least one property, train a support vector machine classifier, and identify an unknown electrical device using the trained support vector machine classifier.
A non-transitory computer-readable storage medium holds instructions that, when executed, enable a computer to identify electrical devices. The instructions cause the computer to capture samples of the alternating current during a device's "ON event," identify the electrical device by analyzing the data in the time and frequency domains, train a support vector machine (SVM) classifier, and identify unknown electrical devices using the trained SVM.
18. The at least one computer readable storage medium of claim 17 , wherein the instructions, if executed, further cause a computing system to: capture a line voltage signal of an alternating current, compute a peak-to-peak voltage of the alternating current, and detect an ON event of the electrical device.
The computer-readable storage medium described previously contains instructions that further cause the computer to capture the line voltage signal of the alternating current, compute the peak-to-peak voltage of the alternating current, and detect the "ON event" of the electrical device. These operations are performed as part of the process of capturing and analyzing the electrical signature for device identification.
19. The at least one computer readable storage medium of claim 17 , wherein the instructions, if executed, further cause a computing system to compute a feature set for the electrical device including frequency domain features and time domain features.
In the described computer-readable storage medium, the instructions further cause the computing system to compute a feature set for each electrical device, including both frequency domain features and time domain features. These features are derived from the captured voltage and current data and are used as input to the support vector machine classifier for device identification.
20. An apparatus, comprising: a sensing module to detect samples corresponding to at least one property of an alternating current of a leg in an electrical system during an ON event; and a computation module, in communication with the sensing module, to identify an electrical device by time domain analysis and frequency domain analysis of the at least one property, wherein the computation module includes: a training module having a training database module and a support vector classifier; and a classification module having a trained support vector machine classifier and a classification database.
An apparatus is designed to identify electrical devices by their electrical characteristics during a power-on event. It comprises a sensing module to detect samples of voltage and current when a device is turned on. A computation module, connected to the sensing module, identifies the device by analyzing this data in both time and frequency domains. The computation module includes a training module, containing a database and support vector machine (SVM) classifier, and a classification module, containing the trained SVM and a database.
21. The apparatus of claim 20 , wherein the sensing module includes: a data acquisition module, a peak detect filtering module, and an event detection module.
In the electrical device identification apparatus, the sensing module is further defined to include a data acquisition module, a peak detect filtering module, and an event detection module. These components work together to capture and prepare the electrical signal for analysis by the computation module. The data acquisition module captures the raw signal, the peak detect filtering module extracts voltage information, and the event detection module identifies when a device turns on.
22. The apparatus of claim 21 , wherein: the data acquisition module is to capture a line voltage signal of an alternating current and perform an analog-to-digital conversion of the signal, the peak detect filtering module is to compute a peak-to-peak voltage of the alternating current, and the event detection module is to detect the ON event of the electrical device.
In the electrical device identification apparatus, the data acquisition module captures the line voltage signal and performs an analog-to-digital conversion. The peak detect filtering module computes the peak-to-peak voltage of the alternating current. The event detection module detects the ON event of the electrical device, triggering the data capture and analysis process. These modules work together to prepare the electrical signal for analysis.
23. The apparatus of claim 20 , wherein the computation module further includes a feature extraction module.
In the electrical device identification apparatus, the computation module further includes a feature extraction module. This module is responsible for extracting relevant features from the captured electrical signal, such as frequency domain features and time domain features, which are then used as input to the support vector machine classifier for device identification.
24. The apparatus of claim 23 , wherein: the feature extraction module is to compute a feature vector including frequency domain features and time domain features, the training module is to train the support vector machine classifier, and the classification module is to identify an unknown electrical device using the trained support vector machine classifier.
In the electrical device identification apparatus, the feature extraction module computes a feature vector that contains both frequency domain features and time domain features. The training module trains the support vector machine (SVM) classifier using these features from known devices. Finally, the classification module identifies an unknown electrical device using the trained SVM classifier based on its extracted features.
Cooperative Patent Classification codes for this invention. Click any code to explore related patents in that topic.
December 28, 2013
March 28, 2017
Browse 5M+ US patents with plain-English claim translations and AI-generated analysis.